WO2014183126A1 - Système et procédé permettant une commutation photonique - Google Patents

Système et procédé permettant une commutation photonique Download PDF

Info

Publication number
WO2014183126A1
WO2014183126A1 PCT/US2014/037724 US2014037724W WO2014183126A1 WO 2014183126 A1 WO2014183126 A1 WO 2014183126A1 US 2014037724 W US2014037724 W US 2014037724W WO 2014183126 A1 WO2014183126 A1 WO 2014183126A1
Authority
WO
WIPO (PCT)
Prior art keywords
photonic
switch
packet
output port
switching fabric
Prior art date
Application number
PCT/US2014/037724
Other languages
English (en)
Inventor
Hamid Mehrvar
Eric Bernier
Peter Ashwood-Smith
Original Assignee
Huawei Technologies Co., Ltd.
Futurewei Technologies, Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huawei Technologies Co., Ltd., Futurewei Technologies, Inc. filed Critical Huawei Technologies Co., Ltd.
Priority to KR1020157035074A priority Critical patent/KR20160006766A/ko
Priority to CN201480024668.8A priority patent/CN105210316B/zh
Priority to EP14794695.8A priority patent/EP2995023B1/fr
Publication of WO2014183126A1 publication Critical patent/WO2014183126A1/fr

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q11/00Selecting arrangements for multiplex systems
    • H04Q11/0001Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005Switch and router aspects
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L49/00Packet switching elements
    • H04L49/35Switches specially adapted for specific applications
    • H04L49/356Switches specially adapted for specific applications for storage area networks
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q11/00Selecting arrangements for multiplex systems
    • H04Q11/0001Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0062Network aspects
    • H04Q11/0066Provisions for optical burst or packet networks
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q11/00Selecting arrangements for multiplex systems
    • H04Q11/0001Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005Switch and router aspects
    • H04Q2011/0007Construction
    • H04Q2011/0033Construction using time division switching
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q11/00Selecting arrangements for multiplex systems
    • H04Q11/0001Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005Switch and router aspects
    • H04Q2011/0037Operation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q11/00Selecting arrangements for multiplex systems
    • H04Q11/0001Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005Switch and router aspects
    • H04Q2011/0037Operation
    • H04Q2011/0045Synchronisation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04QSELECTING
    • H04Q11/00Selecting arrangements for multiplex systems
    • H04Q11/0001Selecting arrangements for multiplex systems using optical switching
    • H04Q11/0005Switch and router aspects
    • H04Q2011/0037Operation
    • H04Q2011/005Arbitration and scheduling

Definitions

  • the present invention relates to a system and method for optical communications, and, in particular, to a system and method for photonic switching.
  • the racks of servers, storage, and input-output functions contain top of rack (TOR) switches which combine packet streams from their associated servers and/or other peripherals into a smaller number of high speed streams per TOR switch which are routed to the packet switching core. Also, TOR switches receive the returning switched streams from that resource and distribute them to servers within their rack. There may be 4 x 40 Gb/s streams from each TOR switch to the packet switching core, and the same number of return streams. There may be one TOR switch per rack, with hundreds to tens of thousands of racks, and hence hundreds to tens of thousands of TOR switches in a data center.
  • TOR top of rack
  • An embodiment method of photonic packet switching includes receiving, by a photonic switching fabric from a first top-of-rack (TOR) switch, a destination port request corresponding to a first photonic packet and a first period of time, where the destination port request includes a first output port and determining whether the first output port is available during the first period of time.
  • the method also includes receiving, by the photonic switching fabric from the first TOR switch, the first photonic packet and routing the first photonic packet to the first output port when the first output port is available during the first period of time. Additionally, the method includes routing the first photonic packet to an alternative output port when the first output port is not available during the first period of time.
  • An embodiment photonic switching fabric includes a first photonic switch and a switch controller coupled to the first photonic switch, wherein the first photonic switch is configured to be coupled to a plurality of top-of-rack (TOR) switches, where the first photonic switch includes a first plurality of input ports and a second plurality of output ports, where the second plurality of output ports is greater than the first plurality of input ports, where the switch controller is configured to determine whether a first output port is available during a first period of time in accordance with a first destination port request, where the first photonic switch is configured to connect a first input port to the first output port when the first output port is available during the first period of time, and where the first photonic switch is configured to connect the first input port to a second output port when the first output port is not available during the first period of time.
  • TOR top-of-rack
  • An embodiment photonic switching fabric includes a photonic packet switch and a plurality of input photonic switches including a first input photonic switch, where the plurality of input photonic switches is coupled to the photonic packet switch, where the plurality of input photonic switches is configured to be coupled to a plurality of top-of rack (TOR) switches including a first TOR switch, where the first input switch is configured to direct a first packet from the first TOR switch to the photonic packet switch when a first output port of the photonic packet switch is available during a first period of time, and where the first input photonic switch is configured to return the photonic packet to the first TOR switch when the first output port of the photonic switching fabric is not available during the first period of time.
  • TOR top-of rack
  • Figure 1 illustrates an embodiment data center
  • Figure 2 illustrates an embodiment star architecture
  • Figure 3 illustrates an embodiment ring architecture
  • Figure 4 illustrates incoming contending photonic packets to a photonic packet switch
  • Figure 5 illustrates a resolution of contended packets by a photonic packet switch
  • Figure 6 illustrates an embodiment photonic switching fabric for contention resolution
  • Figure 7 illustrates another embodiment photonic switching fabric with contention resolution
  • Figure 8 illustrates a flowchart of an embodiment method of resolving contending photonic packets
  • Figure 9 illustrates an additional embodiment photonic switching fabric for contention resolution
  • Figure 10 illustrates a graph of throughput versus switch size
  • Figure 11 illustrates output load balancing in a photonic switching fabric
  • Figure 12 illustrates input load balancing in a photonic switching fabric
  • Figure 13 illustrates another embodiment photonic switching fabric for contention resolution
  • Figure 14 illustrates a graph of the number of packets a switch buffers for each output as a function of interface load
  • Figure 15 illustrates an embodiment photonic network controlled by software defined networking (SDN);
  • Figure 16 illustrates an embodiment photonic switching architecture for contention resolution;
  • Figure 17 illustrates another embodiment photonic switching architecture for contention resolution
  • Figure 18 illustrates an additional embodiment photonic switching architecture for contention resolution
  • Figure 19 illustrates a flowchart for an embodiment method of resolving contending photonic packets
  • Figure 20 illustrates an embodiment system for hybrid photonic packet switching
  • Figures 21A-C illustrate waveforms and an eye diagram for a photonic packet switching system
  • Figure 22 illustrates another embodiment system for photonic packet switching.
  • Corresponding numerals and symbols in the different figures generally refer to corresponding parts unless otherwise indicated.
  • the figures are drawn to clearly illustrate the relevant aspects of the embodiments and are not necessarily drawn to scale.
  • contention resolution One challenge in realizing photonic packet switching is contention resolution.
  • a contention occurs when two or more packets attempt the same output port of a switch at the same time.
  • One method of contention resolution is deflection routing. In deflection routing, contention is resolved by routing only one of the contending packets to the desired output port, while other contended packets are deflected to another path within the node or the network.
  • Contention resolution in a pure photonic packet switching may involve asynchronous contention resolution.
  • contended packets are returned to the source.
  • multiple packets destined for an output port have priority allowing examining the highest priority available port first.
  • FIG. 1 illustrates data center 100, a three tier data center.
  • Server banks 104 contain N servers 102 each. Servers of the server banks are connected to top of rack (TOR) switches 106, the smallest switches. Servers 102 and TOR switches 106 are organized in racks.
  • TOR groups 108, cluster switches, are connected to TOR switches 106. There are M TOR switches per TOR group, and P TOR groups.
  • Traffic is routed from source servers of servers 102 through TOR switches 106, TOR groups 108 to be switched by photonic switch 110.
  • traffic is routed through photonic switch 112 for multihop switching.
  • Data center 100 may include one photonic switch for single hop switching, two photonic switches as pictured in Figure 1, or more photonic switches.
  • Photonic switches 110 and 112 are space switches. In one example, photonic switches 110 and 112 are silicon photonic switches. The traffic then proceeds through TOR groups 108 and TOR switches 106 to destination servers of servers 102.
  • Photonic switching fabric 122 connects TORs or TOR groups 126 in a star configuration. TORs or TOR groups 126 are connected to subnetworks 124. TOR or TOR group 126 may wrap one or many packets destined to another TOR or TOR group into a photonic frame and send the frame to photonic switch 122.
  • the photonic frame has a wavelength encoded label to represent destination TOR or TOR group. More details on photonic frame wrapping are discussed in U.S. Patent Application HW
  • TOR switches 134 are connected to subnetworks 136.
  • high capacity photonic ring 132 has a bandwidth of 1.28 Tbps.
  • stacks of many rings with arbitrary rates are used.
  • the control signals and data use separate wavebands.
  • a number of wavelengths in the signaling waveband may be used for TOR or TOR group addressing.
  • the signaling waveband carries both routing and management information. Because the number of nodes in a ring is limited, a limited number of wavelengths may be used for addressing the TOR groups. For example, out of 12 wavelengths in the 1550 nm range, some may be used to address ring nodes and some are used for management and control.
  • the signaling waveband may also carry other control signals, such as congestion status, fairness, and management.
  • Packet 230 arrives at packet switch 222 before packet 226. As shown in Figure 5, packet 230 is routed to output 3. When packet 226 reaches photonic switch 222, output port 3 is occupied, and packet 226 is returned to the source. The source may again attempt to transmit packet 226 to output port 3. The packet may be retried a number of times. If the packet transmission is not ultimately successful because the there is an extreme contention for the desired output port, the packet may be dropped. Hence, buffering occurs at the source in the electronic domain, instead of at the photonic switch in the optical domain.
  • FIG 6 illustrates photonic packet switch 242.
  • Packet switch contains N input ports to receive packets from source TOR or TOR groups, N return output ports to return contended packets to the source TOR switches, and N output ports to transmit packets to the destination TOR switches. The additional links to return the photonic packets to the source TOR or TOR group are referred to as contention links.
  • photonic packet switch 242 has N input ports and 2N output ports.
  • Figure 7 illustrates photonic switching system 160.
  • Input TOR switches 162 are coupled to photonic switching fabric 166.
  • input TOR switches 162 transmit optical packets to input switches 170.
  • Input switches 170 are 1 :2 photonic switches which route the photonic packets either to photonic switch 172 or back to the source TOR switches.
  • Switch contention control 168 coordinates input switches 170. When a packet label (or header) is received by an input switch, it consults switch contention control on whether the destination port is available. When the destination port is available, the packet is routed to photonic switch 172. The packet traverses the switching elements within the photonic switch 172, set up by switch controller, to reach the output that goes to destination TOR switches 164. When the output port is not available, the photonic packet is switched back to the source TOR.
  • Photonic switch 172 is a 32x32 photonic switch. In one example, photonic switch 172 is a silicon photonic switch.
  • Figure 8 illustrates flowchart 250 for a method of photonic packet switching.
  • the photonic switching fabric receives a photonic packet label from a source, for example from a source TOR switch.
  • the received photonic packet is destined for a particular output port.
  • step 254 the photonic switching fabric determines whether the destination output port is available.
  • the destination output port is unavailable when there is a photonic packet being routed to that output port at the time being requested.
  • the photonic switching fabric proceeds to step 256, and when the destination output port is available, the photonic switching fabric proceeds to step 258.
  • step 256 the photonic switching fabric returns the photonic packet to its source TOR switch.
  • An additional output port is used to route the packet back to the source TOR switch.
  • the source TOR switch may again attempt to transmit the packet to the destination output port.
  • the photonic packet is routed to the requested output of the photonic switch.
  • a 2:1 photonic switch may be used to route the packet either to the photonic switch or back to the source TOR switch.
  • the photonic switching fabric may be an NxN buffer-less optical space switch.
  • step 262 the photonic packet is sent to the destination TOR switch through the path established by the switch controller by examining the label information.
  • photonic switching fabric contending photonic packets are routed to one of several output ports. Load balancing may be used.
  • Figure 9 illustrates photonic switching system 180 for resolving contention using deflection to a dilated part of the fabric.
  • Input TOR switches 182 transmit photonic packets to photonic switching fabric 184.
  • Load balancing is performed by load balancing block 198. Load balancing equally distributes the traffic load to the output ports preventing or reducing packet loss. When load balancing is effective, the inputs and outputs have a similar traffic distribution.
  • a header is sent by the source TOR switches in advance of the packet.
  • the header indicates the destination address to be routed through any of the three choices of destination port.
  • the label is sent to either of the two input ports and read by label detectors 188.
  • the destination address is wavelength encoded, where each wavelength indicates a bit for the destination address.
  • the wavelengths have two power levels. Low power may represent a 0 and high power a 1, or vice versa. More details on wavelength encoding are discussed in U.S. Patent Application Serial No. 13/902,085 filed on May 24, 2013, and entitled "System and Method for Multi-Wavelength Encoding," which application is hereby incorporated herein by reference.
  • Switch controller 190 performs contention analysis and scheduling. In one example, it decides which of the three output ports are available and selects a photonic switch to connect the input on which the packet arrives to that output port. It is possible, but unlikely, given the existence of load balancer with an appropriate dilation level, that none of the choices are available. In this case, the packet is lost.
  • photonic switching fabric 184 When a packet is received by photonic switching fabric 184, it is routed by photonic switches 186, 2x3 photonic switches which route the packet to the appropriate input of photonic switch 192, photonic switch 194, or photonic switch 196. The photonic packet is then switched to the appropriate output port, and sent to output TOR switches 185. The packet is switched based on the decision by switch controller 190 connected to the switching cells in the connection path.
  • photonic packet switch 184 has N input ports and 1.5 N output ports. Each TOR sends out packet on two links and receives packets on 3 links. In one example, there are N input ports and 1.4 N output ports for photonic switching fabric 184.
  • the probability of n of the N packets contending for the same output at the same time is the probability of n of the N packets arriving simultaneously for the same ility is given by:
  • the throughput of the system T is given by the sum of all k probabilities that at least one packet is destined to output k divided by N, given by:
  • Figure 10 illustrates a graph of throughput 400 as a function of N.
  • 37% of the packets will be lost, because there is no buffer to absorb them. This means that adding 40% additional links may absorb these contended packets.
  • the number of output links is 1.35, 1.37, 1.4, 1.5, 1.67, 1.75 or 2 times the number of input links.
  • Figure 11 illustrates output load balancing optical packet switch 202 with one input link which handles ⁇ traffic and N outputs which handle ⁇ / ⁇ traffic each. In load balancing, traffic to all destinations is uniformly distributed.
  • Figure 12 illustrates input load balancing optical packet switch 214 with N inputs and one output.
  • the N inputs each handle ⁇ / ⁇ traffic, while the output handles ⁇ traffic.
  • FIG 13 illustrates photonic switching system 410, which has more dilation and less dependency on load balancing than photonic switching system 180. There may be load balancing.
  • Photonic switching fabric 414 contains N input ports and 2N output ports. Input TOR switches 412 are coupled to photonic switching fabric 414.
  • the buffering requirement may be calculated when there is a uniform distribution of load across all outputs. There may be a maximum number of packets sent to the contention links. A calculation for the buffer requirement may use the Poisson distribution. Although this distribution may underestimate the buffer size for data networks, even this
  • the blocking probability can be obtained using M/M/l/K system.
  • the first M represents the Poisson distribution of arrival of a packet or wrap
  • the second M represents transmission time (or service time) of a packet or wrap to the output port
  • 1 represent the number of switch links to the output destination
  • K represents the number of packets or wraps that can be held for each output port.
  • Headers are sent in advance of the photonic packets.
  • the headers contain four choices of output port for the photonic packet. These four choices may have equal priority and may be represented by a single table, or may have different priorities assigned by a network controller.
  • the headers are read by label detectors 422.
  • the header is wavelength encoded, where the presence or absence of power on a wavelength indicates one bit of the destination addresses.
  • the decoded addresses are sent to switch controller 418.
  • Switch controller 418 determines which of the requested output ports are available. When the output ports have equal priority, any of the output can be assigned. When the output ports have different priorities, the highest priority output port that is available is granted to the photonic packet. Some packets may be dropped, but the probability of a packet being dropped is low.
  • the packet is received by photonic switching fabric 414, it is routed by one of 2x4 input switches 420. These switches determine which input the packet is switched to.
  • photonic switches 426, 428, 424, and 430 switch the photonic packet.
  • These photonic switches may be silicon photonic space switches.
  • Switch controller 418 configures the photonic switches to route the photonic packets.
  • the switched photonic packets are output to destination TOR switches 416.
  • Figure 15 illustrates system 351, a memory-less optical data plane with software defined networking (SDN) for photonic frame switching with many bufferless photonic switches.
  • System 351 has an edge buffer architecture.
  • TORs 359 are coupled to access networks 355.
  • TORs 358 pass packets to photonic switching core 357 for switching.
  • SDN controller 353 is used for source based routing.
  • SDN controller 353 facilitates programmable control of photonic packet switching without physical access to the photonic switches, facilitating source based routing.
  • SDN controller 353 oversees network level routing.
  • Photonic switching core 357 contains wrappers 361, which wrap packets to produce wrapped photonic frames. Wrappers 361 remove the IPG between packets and concatenate the packets, creating gaps between photonic frames. The gap may be about equal to the sum of the removed IPGs.
  • node level routing uses a contention and load balancer.
  • This embodiment has edge photonic switching devices which interface between the electronic access network and the photonic core switches.
  • Edge photonic switches may use a wrapper scheme to send the packet as photonic frames.
  • Photonic switching fabric 442 is an Nxl.4N photonic switching fabric. In another example, photonic switching fabric 442 is an Nxl.5N photonic switching fabric. Photonic switching fabric is coupled to TOR switches 444, which are coupled in a ring or daisy chain configuration. The ring capacity is equal to the extra dilation assumed in the photonic switch. In one example, the ring is a passive ring equipped with a splitter and amplifier. The deflected packet header may be examined by all ring nodes (TOR switches or TOR groups) and collected by the node if the packet destination address matches the node address or an address that belongs to the network attached to this node.
  • TOR switches 444 which are coupled in a ring or daisy chain configuration.
  • the ring capacity is equal to the extra dilation assumed in the photonic switch.
  • the ring is a passive ring equipped with a splitter and amplifier.
  • the deflected packet header may be examined by all ring nodes (TOR switches or TOR groups) and
  • Packets are directed to their destination by deflection to the photonic bufferless ring.
  • Figure 17 illustrates system 450, another system for routing contended packets photonic packets in a photonic switching fabric.
  • photonic switching fabric 452 There are multiple interfaces between photonic switching fabric 452, an NxN photonic switching fabric, and TOR switches 454 and 458.
  • Load balancer 456 balances the loads between TOR switches 454 and TOR switches 458.
  • Load balancer 456 monitors the load of the links to determine if there is an overload situation for a TOR switch or a TOR group. If there is an overload situation, load balancer 456 makes a traffic balancing action to eliminate or minimize packet loss due to a contention that cannot be absorbed by the 40% additional links.
  • the load balancer is optional.
  • FIG 18 illustrates system 140 for routing contended photonic packets with a second switch to handle the additional traffic.
  • Photonic switching fabric 146 an NxN photonic switching fabric, is coupled to TOR switches 142 and 144.
  • Load balancer 150 balances the loads among the links.
  • Photonic packet switching fabric 146 receives photonic packets from the TOR switches. When the requested output port is not available, photonic switching fabric 146 routes the packet to switching fabric 148, which routes the contended packets.
  • switching fabric 148 is a smaller photonic switch.
  • switching fabric 148 is an electrical switch, and may contain buffering.
  • FIG 19 illustrates flowchart 460 for a method of resolving contended photonic packets using deflection.
  • the photonic switching fabric performs load balancing.
  • load When one or more link(s) have a disproportionately high amount of traffic, the load is balanced so that traffic is transferred from overutilized links to underutilized links.
  • Load balancing reduces the rate of dropped packets.
  • the load balancer balances the data storage on the servers connected to TOR switches or TOR groups, so the traffic load on the output links of the photonic switch are uniformly distributed. Given that the data is stored as chunks in a data center with copies of each chunk in many locations, a method of load balancing involves fetching the data from a server connected to a TOR resulting in a balanced demand across switch outputs.
  • the photonic switching fabric receives a destination port request corresponding to a photonic packet to be switched.
  • the destination port request may be in the form of a header.
  • the header is wavelength encoded.
  • the destination port request indicates which switching fabric and which output port(s) of the photonic packet switch has been requested.
  • the destination port request indicates selection of an available output link between photonic packet switch and the destination.
  • the destination port request requests two, three, four, or more output ports going to the same destination.
  • the output ports may be requested in their order of priority.
  • step 466 the photonic switching fabric determines whether the requested destination port is available at the requested time.
  • the port is unavailable when it is being used to switch another photonic packet at that time.
  • the photonic switching fabric proceeds to step 474, and when the requested port is unavailable, the photonic switching fabric proceeds to step 478.
  • all requested destination ports have the same priority.
  • step 478 the photonic switching fabric determines if there is another output port to consider. When there are no output ports to be considered, the packet may be dropped in step 476. When the output port requests are prioritized, the photonic switching fabric considers the next requested output port in step 466.
  • the photonic switch receives the photonic packet to be switched.
  • the photonic packet is received from a TOR switch over an optical fiber.
  • the photonic packet is optically switched.
  • the photonic packet is first switched by a 2x3 or 2x4 photonic switch to direct the photonic packet to the switching fabric which has been scheduled to connect the input to the desired output. Then, the photonic packet is switched by the switching fabric.
  • the photonic packet switch is an optical space switch.
  • the photonic packet is received by a large photonic switch. When the requested output port is available, the photonic switch is switched by the large photonic switch. When the requested output port is unavailable, the photonic packet is sent to a smaller switch which handles the overflow. The photonic packet is then switched by the smaller switch, which may be an optical switch or an electrical switch. Alternatively, only one switch is used.
  • step 472 the switched photonic packet is transmitted.
  • the switched photonic packet is transmitted to a destination TOR switch along an optical fiber.
  • An embodiment photonic switching fabric asynchronously deflects photonic packets without the use of an optical buffer.
  • An embodiment uses optical space switches instead of arrayed waveguide gratings (AWG). Single hop or multi-hop bufferless photonic space switches may be used. Buffering is performed at the TOR switches in the electrical domain, not in the photonic switching fabric in the optical domain.
  • AMG arrayed waveguide gratings
  • Figure 20 illustrates embodiment system 270 with electrical switch 298 and photonic switch 306, where short packets are switched by electrical switch 298 and long packets are switched by photonic switch 306.
  • An embodiment system separates short packets from long packets. Contention resolution can be applied for the photonic switch of system 270. Additional details on a packet switching system where short packets are switched by an electrical switch and fast packets are switched by a photonic switch are discussed in U.S. Patent Application Serial No. 13/902,008 filed on May 24, 2013, and entitled "System and Method for Steering Packet Streams," which application is hereby incorporated herein by reference.
  • Servers 272 and 274 are coupled to TOR switch 280, while servers 276 and 278 are coupled to TOR switch 282.
  • the optical signals for TOR switch 280 and TOR switch 282 are converted to the electrical domain by optical-to-electrical converters 284 and 286, respectively.
  • Processor 288, a field-programmable gate array (FPGA) that may be part of TOR switches 280 and 282, processes the packets. Incoming packets are processed by ingress 291 and ingress 294, while outgoing packets are processed by egress 292 and egress 296.
  • the links between TOR switches 280 and 282 and processor 288 are 10 Gigabit Ethernet.
  • a label is followed by a scrambled media access control (MAC) frame.
  • MAC media access control
  • IP internet protocol
  • the processed packets are then converted from the electrical domain to the optical domain by electrical-to-optical converters 290, 293, 295, and 297.
  • Short packets are routed to electrical-to-optical converters 290 and 295 and proceed to be switched by electrical switch 298.
  • Photonic switch 306 Long packets are routed to photonic switch 306, a 4x4 Lead-Lanthanum-Zirconate- Titanate (PLZT) photonic switch.
  • the switching time of photonic switch 306 is about 10-20 ns.
  • Fiber splitters 301 direct 10% of the power to optical-to-electrical converters 302.
  • the electrical signals are used to control photonic switch 306 by switch controller 304, an FPGA based switch controller.
  • Fiber delay lines 303 delay the signal long enough for the switch controller to read the photonic label and set the switch connection before the packet arrives.
  • Figures 21A-C illustrate results from system 270 in Figure 20.
  • Server 272 sends Ethernet packets with four different destination MAC addresses, each destined to a different photonic output port of photonic switch 306.
  • Figure 21A illustrates graph 310 with the packet waveform on the four output ports of photonic switch 306. The photo-receiver voltage polarity is inverted, with horizontal lines when there is no light and the waveforms when there are switched packets.
  • Figure 21B illustrates graph 320 with a detailed output packet waveform of output ports 1 and 2 of photonic switch 306.
  • Output 1 completes a photonic frame transmission, and output 2 starts sending a preamble and photonic label.
  • Switch response time is 12 ns
  • residual preamble for receiver synchronization is 15 ns
  • start frame delimiter (SFD) time is 12 ns.
  • Figure 21 C illustrates graph 330 with an eye diagram of the switched signal.
  • Figure 22 illustrates system 340, an embodiment photonic switching system that uses optical space switching.
  • System 340 may be an implementation of photonic switching system 160 in Figure 7. Separate wavebands are used for the control signal path and the payload data path. Photonic routing labels are used on the forward path. Signaling on the return path is used for contention control and synchronization.
  • System 340 uses an embodiment by deploying 2 inputs of the photonic switch and 4 outputs.
  • the additional outputs carry the contented packets.
  • Server network 342 is simulated by simulator 344 and simulator 346.
  • Simulators 344 and 346 contain small form factor pluggable transceivers (SFPs) 348, 350, 352, and 354, which are connected to TOR switches 356, 358, 360, and 362. The signals are sent to FPGA 366.
  • SFPs small form factor pluggable transceivers
  • FPGA 366 signals are received by SFP 368. These signals are proceed by front- end adaptor 372. Labels are generated by label generator 374. The signals and groups are output by SFP 378 to photonic switching fabric 386 and FPGA 390.
  • the optical signal of the labels is converted to an electrical signal by optical-to- electrical converters 398, and is received by FPGA 390. They are processed by processor 396. Then, the control signal is extracted by control signal extractor 394. The control signals are then converted by low- voltage differential signal (LVDS) to transistor-transistor logic (TTL) board 392.
  • LVDS low- voltage differential signal
  • TTL transistor-transistor logic
  • the data wave path signals and the signaling wave path signals are multiplexed by multiplexer 380, with data at 40GE and signaling at 10GE, and output to photonic switching fabric 386.
  • the control signals from FPGA 390 are also input to photonic switching fabric 386.
  • Photonic switching fabric 386 is a 4x4 optical space switch. The signals are switched, and output to FPGA 366.
  • the signals are received by demultiplexer 382 and SFP 378. They are processed by back-end adaptor 376. The signals are converted by FPGA mezzanine card (FMC) to subminiature version A (SMA) converter 370. The signals are converted to optical signals by electrical-to-optical converters 364, and proceed to TOR switches 356, 358, 360, and 362.
  • FMC FPGA mezzanine card
  • SMA subminiature version A

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Optical Communication System (AREA)

Abstract

La présente invention se rapporte, selon un mode de réalisation, à un procédé permettant une commutation de paquets photoniques, ledit procédé consistant à recevoir, au moyen d'une matrice de commutation photonique d'un premier commutateur de haut de bâti (TOR pour Top-Of-Rack), une requête de port de destination correspondant à un premier paquet photonique et une première période de temps, la requête de port de destination comprenant un premier port de sortie, et à déterminer si le premier port de sortie est disponible pendant la première période de temps. Le procédé consiste également à recevoir, au moyen de la matrice de commutation photonique du premier commutateur de haut de bâti (TOR), le premier paquet photonique et à acheminer le premier paquet photonique jusqu'au premier port de sortie lorsque le premier port de sortie est disponible pendant la première période de temps. De plus, le procédé consiste à acheminer le premier paquet photonique jusqu'à un port de sortie alternatif lorsque le premier port de sortie n'est pas disponible.
PCT/US2014/037724 2013-05-10 2014-05-12 Système et procédé permettant une commutation photonique WO2014183126A1 (fr)

Priority Applications (3)

Application Number Priority Date Filing Date Title
KR1020157035074A KR20160006766A (ko) 2013-05-10 2014-05-12 광 스위칭을 위한 시스템 및 방법
CN201480024668.8A CN105210316B (zh) 2013-05-10 2014-05-12 用于光子交换的系统和方法
EP14794695.8A EP2995023B1 (fr) 2013-05-10 2014-05-12 Système et procédé permettant une commutation photonique

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
US201361822180P 2013-05-10 2013-05-10
US61/822,180 2013-05-10
US14/275,320 US9661405B2 (en) 2013-05-10 2014-05-12 System and method for photonic switching
US14/275,320 2014-05-12

Publications (1)

Publication Number Publication Date
WO2014183126A1 true WO2014183126A1 (fr) 2014-11-13

Family

ID=51867800

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2014/037724 WO2014183126A1 (fr) 2013-05-10 2014-05-12 Système et procédé permettant une commutation photonique

Country Status (3)

Country Link
US (1) US9661405B2 (fr)
KR (1) KR20160006766A (fr)
WO (1) WO2014183126A1 (fr)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016119577A1 (fr) * 2015-01-29 2016-08-04 Huawei Technologies Co., Ltd. Systèmes et procédés de programmation de paquets dans un système de commutation optique
WO2016180357A1 (fr) * 2015-05-14 2016-11-17 Huawei Technologies Co., Ltd. Système et procédé de commutation photonique
WO2016184345A1 (fr) * 2015-05-15 2016-11-24 Huawei Technologies Co., Ltd. Système et procédé de commutation photonique
WO2017050099A1 (fr) * 2015-09-25 2017-03-30 Huawei Technologies Co., Ltd. Commutation d'interface de matrice et ordonnancement
US11316796B2 (en) 2019-12-30 2022-04-26 Juniper Networks, Inc. Spraying for unequal link connections in an internal switch fabric

Families Citing this family (34)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9432748B2 (en) 2013-05-10 2016-08-30 Huawei Technologies Co., Ltd. System and method for photonic switching
US9560427B2 (en) * 2014-05-28 2017-01-31 Huawei Technologies Co., Ltd. Scalable silicon photonic switching architectures for optical networks
CN105282629A (zh) * 2014-07-03 2016-01-27 中兴通讯股份有限公司 一种硅光交叉连接的控制方法和装置
US9860614B2 (en) * 2015-05-13 2018-01-02 Huawei Technologies Co., Ltd. System and method for hybrid photonic electronic switching
CN106817288B (zh) * 2015-11-30 2019-06-14 华为技术有限公司 一种数据中心网络系统及信号传输系统
US10750255B2 (en) * 2016-04-22 2020-08-18 Huawei Technologies Co., Ltd. Segment routing for optical networks
US10306344B2 (en) * 2016-07-04 2019-05-28 Huawei Technologies Co., Ltd. Method and system for distributed control of large photonic switched networks
KR20180049401A (ko) 2016-11-01 2018-05-11 주식회사 아모그린텍 전극 및 이를 이용한 이차전지와 전극의 제조방법
US10499125B2 (en) * 2016-12-14 2019-12-03 Chin-Tau Lea TASA: a TDM ASA-based optical packet switch
US10917707B2 (en) * 2017-01-19 2021-02-09 Telefonaktiebolaget Lm Ericsson (Publ) Network and method for a data center
US10574580B2 (en) 2017-07-04 2020-02-25 Vmware, Inc. Network resource management for hyper-converged infrastructures
US11251878B2 (en) 2018-02-07 2022-02-15 Infinera Corporation Independently routable digital subcarriers for optical communication networks
US11368228B2 (en) 2018-04-13 2022-06-21 Infinera Corporation Apparatuses and methods for digital subcarrier parameter modifications for optical communication networks
CN112042132A (zh) * 2018-04-13 2020-12-04 康普技术有限责任公司 可配置广域分布式天线系统
US11095389B2 (en) 2018-07-12 2021-08-17 Infiriera Corporation Subcarrier based data center network architecture
US10831572B2 (en) * 2018-11-08 2020-11-10 At&T Intellectual Property I, L.P. Partition and access switching in distributed storage systems
US11075694B2 (en) 2019-03-04 2021-07-27 Infinera Corporation Frequency division multiple access optical subcarriers
US11258528B2 (en) 2019-09-22 2022-02-22 Infinera Corporation Frequency division multiple access optical subcarriers
US11336369B2 (en) 2019-03-22 2022-05-17 Infinera Corporation Framework for handling signal integrity using ASE in optical networks
US11418312B2 (en) 2019-04-19 2022-08-16 Infinera Corporation Synchronization for subcarrier communication
US11838105B2 (en) 2019-05-07 2023-12-05 Infinera Corporation Bidirectional optical communications
US11476966B2 (en) 2019-05-14 2022-10-18 Infinera Corporation Out-of-band communication channel for subcarrier-based optical communication systems
US11239935B2 (en) 2019-05-14 2022-02-01 Infinera Corporation Out-of-band communication channel for subcarrier-based optical communication systems
US11190291B2 (en) 2019-05-14 2021-11-30 Infinera Corporation Out-of-band communication channel for subcarrier-based optical communication systems
US11296812B2 (en) 2019-05-14 2022-04-05 Infinera Corporation Out-of-band communication channel for subcarrier-based optical communication systems
US10965378B2 (en) 2019-05-14 2021-03-30 Infinera Corporation Out-of-band communication channel for sub-carrier-based optical communication systems
US11489613B2 (en) 2019-05-14 2022-11-01 Infinera Corporation Out-of-band communication channel for subcarrier-based optical communication systems
US11483257B2 (en) 2019-09-05 2022-10-25 Infinera Corporation Dynamically switching queueing schemes for network switches
US20210111802A1 (en) 2019-10-10 2021-04-15 Infinera Corporation Hub-leaf laser synchronization
CA3157806A1 (fr) 2019-10-10 2021-04-15 Infinera Corporation Systemes de commutateurs de reseaux pour reseaux de communications optiques
AU2020364088A1 (en) * 2019-10-10 2022-05-12 Infinera Corporation Optical subcarrier dual-path protection and restoration for optical communications networks
US20220394362A1 (en) * 2019-11-15 2022-12-08 The Regents Of The University Of California Methods, systems, and devices for bandwidth steering using photonic devices
GB2597487A (en) * 2020-07-23 2022-02-02 British Telecomm Optical network configuration
WO2022123686A1 (fr) * 2020-12-09 2022-06-16 日本電信電話株式会社 Dispositif d'optimisation de connexion, procédé d'optimisation de connexion et programme

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5091905A (en) * 1988-03-18 1992-02-25 Hitachi, Ltd. High-speed packet switching using a space division optical switch
US20020024700A1 (en) * 2000-08-29 2002-02-28 Kddi Corporation Reflectiion routing method in optical packet switching network and optical packet switch for reflection routing
US20020131675A1 (en) * 2001-03-19 2002-09-19 General Instrument Corporation Time slot tunable all-optical packet data routing switch
US6714552B1 (en) * 1996-08-28 2004-03-30 British Telecommunications Public Limited Company Communications network
US7245829B1 (en) * 2004-01-12 2007-07-17 Nortel Networks Limited Architecture for dynamic connectivity in an edge photonic network architecture
US20100036903A1 (en) * 2008-08-11 2010-02-11 Microsoft Corporation Distributed load balancer
US20120099863A1 (en) * 2010-10-25 2012-04-26 Nec Laboratories America, Inc. Hybrid optical/electrical switching system for data center networks
WO2013049675A1 (fr) * 2011-09-30 2013-04-04 Gigamon Llc Systèmes et procédés pour mettre en œuvre un réseau doté d'outils de visibilité du trafic réseau
US20130108259A1 (en) * 2011-11-01 2013-05-02 Plexxi Inc. Affinity modeling in a data center network

Family Cites Families (38)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2283627C (fr) * 1999-09-27 2008-08-12 Nortel Networks Corporation Commutateur de paquets mrl-mrt a grande capacite
US6735199B1 (en) * 1999-11-09 2004-05-11 Synchrodyne Networks, Inc. Time frame switching responsive to global common time reference
US6487332B1 (en) * 1999-12-23 2002-11-26 Lucent Technologies Inc. Strictly non-blocking wavelength division multiplexed (WDM) cross-connect device for use in a heterogeneous network
US6839322B1 (en) * 2000-02-09 2005-01-04 Nortel Networks Limited Method and system for optical routing of variable-length packet data
US20020018263A1 (en) * 2000-06-08 2002-02-14 An Ge Scalable WDM optical IP router architecture
US6816487B1 (en) * 2000-07-20 2004-11-09 Nortel Networks Limited Mapping of high bandwidth connections in a multi-stage switch
US20020037130A1 (en) * 2000-08-02 2002-03-28 Sarnoff Corporation Microfluidic optical switch
JP3472809B2 (ja) * 2000-09-06 2003-12-02 独立行政法人通信総合研究所 多波長ラベルを用いた光パケットルーティング方法とその装置、および多波長ラベルを用いた光パケットネットワーク
US6920135B1 (en) * 2001-01-23 2005-07-19 Tau Networks Scalable switching system and method
US7599620B2 (en) * 2001-06-01 2009-10-06 Nortel Networks Limited Communications network for a metropolitan area
US6768827B2 (en) * 2002-01-16 2004-07-27 The Regents Of The University Of California Integrated optical router
US7260102B2 (en) * 2002-02-22 2007-08-21 Nortel Networks Limited Traffic switching using multi-dimensional packet classification
US20040037558A1 (en) * 2002-08-20 2004-02-26 Nortel Networks Limited Modular high-capacity switch
US6940863B2 (en) * 2003-01-13 2005-09-06 The Regents Of The University Of California Edge router for optical label switched network
JP4009946B2 (ja) * 2003-01-16 2007-11-21 横河電機株式会社 光経路制御装置
US8064341B2 (en) * 2003-10-10 2011-11-22 Nortel Networks Limited Temporal-spatial burst switching
US7403473B1 (en) * 2003-12-29 2008-07-22 Nortel Networks Limited Method and apparatus for accelerated protection switching in a multi-switch network element
US20060165081A1 (en) * 2005-01-24 2006-07-27 International Business Machines Corporation Deflection-routing and scheduling in a crossbar switch
WO2006080279A1 (fr) * 2005-01-28 2006-08-03 Kabushiki Kaisha Route Lamda Dispositif de transmission de signal optique et réseau optique de communication
US8059647B2 (en) * 2005-10-05 2011-11-15 Nortel Networks Limited Multicast implementation in a link state protocol controlled ethernet network
US20070091828A1 (en) * 2005-10-26 2007-04-26 Nortel Networks Limited Registration, look-up, and routing with flat addresses at enormous scales
WO2007056713A2 (fr) 2005-11-04 2007-05-18 The Trustees Of Columbia University In The City Of New York Reseau optique
US8295700B2 (en) * 2007-05-14 2012-10-23 Intune Technologies Distributed packet switch for use in a network
GB0819616D0 (en) * 2008-10-25 2008-12-03 Ct For Integrated Photonics Th Wavelenghth division multiplexing transmission eqipment
CA2781060C (fr) * 2010-05-28 2016-03-08 Huawei Technologies Co., Ltd. Couche virtuelle 2 et mecanisme pour la rendre evolutive
US9001827B2 (en) * 2010-12-17 2015-04-07 Big Switch Networks, Inc. Methods for configuring network switches
JP5426604B2 (ja) * 2011-04-26 2014-02-26 富士通テレコムネットワークス株式会社 光パケット交換システム
JP5842428B2 (ja) * 2011-07-21 2016-01-13 富士通株式会社 光ネットワークおよび光接続方法
US9100116B2 (en) * 2011-08-24 2015-08-04 Ciena Corporation Short-term optical recovery systems and methods for coherent optical receivers
US9250941B2 (en) * 2011-09-30 2016-02-02 Telefonaktiebolaget L M Ericsson (Publ) Apparatus and method for segregating tenant specific data when using MPLS in openflow-enabled cloud computing
US8560663B2 (en) * 2011-09-30 2013-10-15 Telefonaktiebolaget L M Ericsson (Publ) Using MPLS for virtual private cloud network isolation in openflow-enabled cloud computing
AU2012321038B2 (en) 2011-10-04 2014-10-09 Thales Australia Limited Portable device to control simulated aircraft in air traffic control training system
US9059888B2 (en) * 2012-02-16 2015-06-16 Nec Laboratories America, Inc. MIMO-OFDM-based flexible rate intra-data center network
US8983293B2 (en) * 2012-04-25 2015-03-17 Ciena Corporation Electro-optical switching fabric systems and methods
US8831000B2 (en) * 2012-10-10 2014-09-09 Telefonaktiebolaget L M Ericsson (Publ) IP multicast service join process for MPLS-based virtual private cloud networking
US9628878B2 (en) * 2012-12-11 2017-04-18 Huawei Technologies Co., Ltd. System and method for multi-wavelength encoding
US8971321B2 (en) * 2012-12-11 2015-03-03 Futurewei Technologies, Inc. System and method for accelerating and decelerating packets
US10015111B2 (en) * 2013-03-15 2018-07-03 Huawei Technologies Co., Ltd. System and method for steering packet streams

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5091905A (en) * 1988-03-18 1992-02-25 Hitachi, Ltd. High-speed packet switching using a space division optical switch
US6714552B1 (en) * 1996-08-28 2004-03-30 British Telecommunications Public Limited Company Communications network
US20020024700A1 (en) * 2000-08-29 2002-02-28 Kddi Corporation Reflectiion routing method in optical packet switching network and optical packet switch for reflection routing
US20020131675A1 (en) * 2001-03-19 2002-09-19 General Instrument Corporation Time slot tunable all-optical packet data routing switch
US7245829B1 (en) * 2004-01-12 2007-07-17 Nortel Networks Limited Architecture for dynamic connectivity in an edge photonic network architecture
US20100036903A1 (en) * 2008-08-11 2010-02-11 Microsoft Corporation Distributed load balancer
US20120099863A1 (en) * 2010-10-25 2012-04-26 Nec Laboratories America, Inc. Hybrid optical/electrical switching system for data center networks
WO2013049675A1 (fr) * 2011-09-30 2013-04-04 Gigamon Llc Systèmes et procédés pour mettre en œuvre un réseau doté d'outils de visibilité du trafic réseau
US20130108259A1 (en) * 2011-11-01 2013-05-02 Plexxi Inc. Affinity modeling in a data center network

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016119577A1 (fr) * 2015-01-29 2016-08-04 Huawei Technologies Co., Ltd. Systèmes et procédés de programmation de paquets dans un système de commutation optique
US9614778B2 (en) 2015-01-29 2017-04-04 Huawei Technologies Co., Ltd. Systems and methods for packet scheduling in a photonic switching system
WO2016180357A1 (fr) * 2015-05-14 2016-11-17 Huawei Technologies Co., Ltd. Système et procédé de commutation photonique
US9860615B2 (en) 2015-05-14 2018-01-02 Huawei Technologies Co., Ltd. System and method for photonic switching
WO2016184345A1 (fr) * 2015-05-15 2016-11-24 Huawei Technologies Co., Ltd. Système et procédé de commutation photonique
US9654849B2 (en) 2015-05-15 2017-05-16 Huawei Technologies Co., Ltd. System and method for photonic switching
WO2017050099A1 (fr) * 2015-09-25 2017-03-30 Huawei Technologies Co., Ltd. Commutation d'interface de matrice et ordonnancement
US11316796B2 (en) 2019-12-30 2022-04-26 Juniper Networks, Inc. Spraying for unequal link connections in an internal switch fabric

Also Published As

Publication number Publication date
US20150289035A1 (en) 2015-10-08
KR20160006766A (ko) 2016-01-19
US9661405B2 (en) 2017-05-23

Similar Documents

Publication Publication Date Title
US9661405B2 (en) System and method for photonic switching
US9432748B2 (en) System and method for photonic switching
US9654853B2 (en) System and method for photonic switching
US10506311B2 (en) System and method for optical network
US20180287818A1 (en) Non-blocking any-to-any data center network having multiplexed packet spraying within access node groups
US20180287965A1 (en) Non-blocking any-to-any data center network with packet spraying over multiple alternate data paths
US9584885B2 (en) System and method for photonic switching
US9042380B2 (en) Crossbar switch and recursive scheduling
EP2995023B1 (fr) Système et procédé permettant une commutation photonique
Pattavina Architectures and performance of optical packet switching nodes for IP networks
Huang et al. Modeling and performance analysis of OPS data center network with flow management using express path
Li et al. Performance analysis and experimental demonstration of a novel network architecture using optical burst rings for interpod communications in data centers
Ye et al. Assessment of optical switching in data center networks
Kharroubi et al. Approaches and controllers to solving the contention problem for packet switching networks: a survey
US10129049B2 (en) Data transmission method and media access controller
Mobilon et al. Hardware architecture for optical packet and burst switching applications
Sowailem et al. Contention resolution strategy in optical burst switched datacenters
Yu et al. Enhanced fat tree-an optical/electrical hybrid interconnection for data center
Binh et al. Improved WDM packet switch architectures with output channel grouping
Furukawa et al. Multihomed optical packet and circuit integrated network based on hierarchical address allocation
Xu et al. An AWGR-based bufferless interconnect for data centers
Pattavina Performance of IP optical packet networks with deflection routing
Munoz On the sliding-window packet-switching architecture for large-scale internet routers and switches

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14794695

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 2014794695

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 20157035074

Country of ref document: KR

Kind code of ref document: A